Change provider balance on slow delivery every 5 minutes #4287
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
rparke
approved these changes
Dec 3, 2024
CrystalPea
force-pushed
the
change_provider_weighing_more_often_when_needed
branch
2 times, most recently
from
305bdbc to
029a8bf
Compare
Instead of every 10 minutes - this way we react faster, and fewer text messages get delayed. This becomes more important as we now send lots of login codes for various organisations. Here are some considerations for the change: 1. if we move from 10 minutes to 5 minutes, wouldn’t we use the notifications in our decision that were used in previous weighing too? This is a valid question. This is happening already with current set up tho - as we look for last 15 minutes of sms, but we are changing weighing every 10 minutes. So this wouldn’t be a new issue. Also this can be solved to either prioritise the best possible service for our users, or to prioritise fairest possible outcome for our providers. I believe getting text messages delivered promptly is more important than if a provider briefly loses more traffic because they were slow within last 15 minutes. Also if the slow down was just a short blip, on a subsequent delivery speed check some of the notifications will be from after the blip passed, and that should mean their priority has less chance of reducing further. 2. we should learn more before we merge it On one side, fair, on the other, I kinda think it’s not a fundamental change, but rather a tuning, and we did give it some thought already, no? We identified a problem (we adjust ratios too slowly when there is a period of slow delivery for one of our providers), we got an adjustment that should improve this, and we thought of potential flip-sides above (providers ratio will be lowered more quickly when they are slow, that means we will lower their ratio one more time than we would do before after their delivery speed recovers).
CrystalPea
force-pushed
the
change_provider_weighing_more_often_when_needed
branch
from
029a8bf to
6a422f8
Compare
klssmith
reviewed
Dec 6, 2024
We used to only do it every one hour - let's go down to every 15 minutes, to balance with shorter intervals for taking traffic away on slow delivery.
CrystalPea
force-pushed
the
change_provider_weighing_more_often_when_needed
branch
from
6a422f8 to
c3822c1
Compare
klssmith
approved these changes
Dec 13, 2024
CrystalPea
deleted the
change_provider_weighing_more_often_when_needed
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Instead of every 10 minutes - this way we react faster, and fewer text messages get delayed.
This becomes more important as we now send lots of login codes for various organisations.
Also rebalance back to the middle every 15 minutes, instead of every 1 hour.
Those two changes together make our providers balancing act more dynamic - hopefully it will help us deliver messages quickly when one of the providers has delivery issues, while also giving them traffic back more quickly when they recover.
Check commit messages for more detail.